The purpose of this memorandum is to further elaborate on the definition of the term, deep learning, and discuss how it is defined and used by researchers and industry professionals. I will also be analyzing how the term is used contextually in across a variety of electronic publications. After analyzing how it is defined in other written works and used contextually, I will provide my own working definition of the term, deep learning, in relation to my major, computer system technology (CST).
“Deep learning systems are based on multilayer neural networks and power… Combined with exponentially growing computing power and the massive aggregates of big data, deep-learning neural networks influence the distribution of work between people and machines.” (“Neural Network”, 2020) Although the Encyclopedia Britannica does not directly define the term deep learning, it explains the concept under the term neural network. While researching for definitions of deep learning, it was not uncommon to find deep learning and neural network being used together when defining what a neural network is. This is because deep learning is one of the methods neural networks use when analyzing data. Cho (2014) states that “…deep learning, has gained its popularity recently as a way of learning deep, hierarchical artificial neural networks.” (p. 15) This is further demonstrating the fact that deep learning can be defined as a way of helping neural networks learn deeply from the data it receives. In another instance, De (2020) defines deep learning as “…as a particular type of machine learning that uses artificial neural networks.” (p.353) In this particular definition of deep learning, we are made aware of the fact that deep learning is in fact a subset of machine learning, the ability for a computer to learn from data and adjust itself accordingly with minimal to no user input (“Machine”, 2020), that works in conjunction with neural networks to learn from the data inputted without the need for user intervention.
Now that we have explored a few definitions of the term deep learning, let us take a moment to see how the term is used contextually from a variety of sources. The IBM Corporation created a web page dedicated to explaining what deep learning is and its purpose. On this web page, the IBM Corporation (2020) states that, “Deep learning algorithms perform a task repeatedly and gradually improve the outcome through deep layers that enable progressive learning.” (para. 1) Here we can see how a reputable technology company describes how deep learning is able to learn from itself and improve its ability to interpret data more effectively. As the IBM Corporation stated, it is also important to note that deep learning is a progressive learning process and may take many iterations before it can effectively interpret large amounts of data without the need for user interference. In the ScienceDaily, a website dedicated to providing its visitors with the latest news on scientific discoveries from a variety of industries, we can see how the term deep learning is being used in the scientific research industry. In an article by the Institute of Science and Technology Austria (2020), it states that a group of international researchers from Austria, Vienna, and the USA, have developed a new artificial intelligence system that “…has decisive advantages over previous deep learning models: It copes much better with noisy input, and, because of its simplicity, its mode of operation can be explained in detail.” (para. 1) As the IBM Corporation had mentioned, deep learning is a progressive learning process and, in this case, the researchers mentioned in the article were able to further improve upon the current deep learning models to allow for better interpretation of input data. Chen (2018), a science reporter at The Verge, a multimedia technology news source, posted the transcript of an interview she had with Terrence Sejnowski, a computational neuroscientist at the Salk Institute for Biological Studies, in which he said “Buzzwords like “deep learning” and “neural networks” are everywhere, but so much of the popular understanding is misguided.” (para. 1) It is important to note that there is a lot of hype surrounding machine learning, artificial intelligence and deep learning, and that a lot of the information that is readily available can be misinterpreted or as Sejnowski said “misguided”.
After reviewing the material I used to extract quotes from for the definition and context section of the memo, I will develop my own working definition of what deep learning means to me and it relates to my major, CST. I would define deep learning as an iterative learning method used by computers to interpret data inputted by a user without the assistance of the user.
Chen, A. (2018, October 16). A pioneering scientist explains ‘deep learning’. Retrieved October 26, 2020, from https://www.theverge.com/2018/10/16/17985168/deep-learning-revolution-terrence-sejnowski-artificial-intelligence-technology
Cho, K. (2014). Foundations of advances in deep learning [Doctoral dissertation, Aalto University]. https://aaltodoc.aalto.fi/handle/123456789/12729
De, A., Sarda, A., Gupta, S., & Das, S. (2020). Use of artificial intelligence in dermatology. Indian Journal of Dermatology, 65(5), 352–357. https://doi-org/10.4103/ijd.IJD_418_20
IBM Corporation. (2020, September 30). Deep Learning – Neural Networks and Deep Learning. Retrieved October 26, 2020, from https://www.ibm.com/cloud/deep-learning?p1=Search
Institute of Science and Technology Austria. (2020, October 13). New Deep Learning Models: Fewer Neurons, More Intelligence. Retrieved October 26, 2020, from https://ist.ac.at/en/news/new-deep-learning-models/
Machine. (2020). In OED Online. Retrieved from www.oed.com/view/Entry/111850.
Neural network. (2020). In Encyclopedia Britannica. Retrieved from https://academic-eb-com.citytech.ezproxy.cuny.edu/levels/collegiate/article/neural-network/126495